Parameter Optimisation of e.g. Qualitiy Measure using the Rule Engine

Hello guys!

I am trying to optimise the parameters. The Quality Measure and the Pruning from the Decision Tree Learner Node.

Example for the Quality Measure to test whether the Gini Index or Gain ratio leads to a higher accuracy.

To do this, I have the parameter in the optimisation loop, then I try to define the whole thing in the rule engine, but unfortunately this doesn’t work, do you know why this could be?

I cant select Replace Column and get an error all the time.

(Pictures and workflow included)

Thanks for ur help

Decision Tree Modell Para.knwf (117.8 KB)

If you export your data in resetted state no one can see have a look at your issue because the data is missing.
Quality Measure does neither appear in your column list nor in your flow variable list so you can’t reference that in your filter rules

1 Like

Oh dear, thank goodness I fixed that problem but I have another one now, maybe you can have a look at this?

I always get this error:
Error loading flow variables into node: unable to parse split criterion “1”

Workflow (no reset):
Random Forest - not working.knwf (1.1 MB)

split criterion requires string datatype and you provide a number. You need to change the data type

1 Like

Awesome thanks for ur answer what node do I use for it?

Number to String node can convert numerical column(s) to string.

What am I doing wrong?

Random Forest Modell Para.knwf (1.1 MB)

You need to add the table row to variable after number to string and feed that into the learner node.
Also split criteria requires a string like “Gini” and not a number. Replace the name Quality_Measure in your rule engine with something else e.g. abc and use this column in the variable (split criterion) configuration in the learner.

1 Like

Thank you so much for ur help, I think i did everything u said, sadly i got an new error:

Random Forest Modell Para.knwf (1.1 MB)

Remove the space in your rule engine
Information Gain → InformationGain


Thank you so much u saved me, ur help means a lot to me. :heart:

This topic was automatically closed 7 days after the last reply. New replies are no longer allowed.